Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/3735#discussion_r111978489
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/table/runtime/aggregate/GeneratedAggregations.scala
---
@@ -36,6 +36,15 @@ abstract class GeneratedAggregations extends Function {
def setAggregationResults(accumulators: Row, output: Row)
/**
+ * Calculates the results from accumulators, and set the results to the
output (with key offset)
+ *
+ * @param accumulators the accumulators (saved in a row) which contains
the current
+ * aggregated results
+ * @param output output results collected in a row
+ */
+ def setAggregationResultsWithKeyOffset(accumulators: Row, output: Row)
--- End diff --
I don't think we need to extend the `GeneratedAggregations` interface
(except for `resetAccumulators()`)
I would rather implement another code generation function that implements
the existing methods differently. This would mean to add another method to
`CodeGenerator` that generates the `GeneratedAggregations` interface suitable
for the DataSet aggregations.
- `setAggregationResultsWithKeyOffset` -> `setAggregationResults`
- `setKeyToOutput` -> `setForwardedFields`
- `accumulateWithKeyOffset` -> `accumulate`
- `createAccumulatorsAndSetToOutput` could be replaced by
`createAccumulators` (called once to create a reusable accumulators),
`resetAccumulators`, and `setAggregationResults` (if it sets the accumulators
instead of calling `AggFunction.getValue()`, see below)
- `copyAccumulatorsToBuffer` -> `setAggregationResults` (the accumulators
are partial aggregation results). This would mean we have two behaviors,
setting the final (`getValue()`) or the partial result (accumulator) for
`setAggregateResults()`. A simple flag during code gen would go for either the
final or the partial result.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
